1. Technical Field
Methods and apparatuses for multispectral detection of substances in a given scene or field of view.
2. Description of Related Art
Successful optical detection of specific materials and substances in a scene is dependent on being able to distinguish the signal signature within received light relating to their reflection, absorption or emission characteristics from those of background sources and noise. Commercially available image sensors with integrated patterned filter arrays provide a convenient means of sensing received light and acquiring multiband data that can be fed through algorithms to examine inter-band relationships for detection-indicating criteria. However, the multiple spectral response bands of these sensors are typically optimized for criteria other than detection of specific materials and substances such as reducing visible artifacts like noise and aliasing. Distinct spectral response bands are usually created by placing a micro-scale pattern of pigment or dye-based filters, commonly called a color filter array (CFA), in front of a photo-detector array that on its own has little or no native wavelength selectivity. The addition of the CFA creates a sensor whereby each photo-detector (pixel) is assigned to just one of a plurality of spectral response band classes by the transmission properties of the CFA disposed above it.
The CFA pattern and the spectral transmittances of its filters are usually designed to produce data that can be converted to accurate color images through a process called demosaicing. In this process the observed signals from the two dimensional array of pixels, each having a specific spectral response, typically peaking in the red, green and blue (RGB) portions of the visible spectrum, are used to create a multiband image where spectral responses for all bands are estimated at all pixel locations. The estimation is often done through linear or non-linear interpolation of the sparse spatial sampling of the CFA pattern. Furthermore, by examining pixel values between spectral response bands, a determination can be made as to whether the light from the scene likely contains a portion of reflected, absorbed or emitted by the material or substance that is to be detected. Methods for examining the intra-band relationships to determine if the material or substance is present require that the observed signal can be distinguished from signals from other sources such as reflections from materials or substances not of interest (background) and noise.